rburghol
commented
6 years ago The most important step for this will be visualization to describe the anomalies as well as the “non-anomalies”. These visualizations need to be really high quality, with as much narrative detail as you can provide to underscore what conclusions you make. And ultimately, I DO want you to make conclusions, well reasoned, data driven conclusions (this is super-important as our calibration evaluation depends on it) — with that as the goal, here are the requirements I see, and for everyones viewing ease, please put them here in this github thread (copy and paste images works on chrome):
- let’s produce full drainage maps for all gages as part of the documentation package (if you have those then please resend - sorry if I missed it)
- clearly show the outline of the full drainage area of the model segment in question — we’re now in the phase where we need too see individually what each are doing
- exploration of the next downstream segment (and visualization) in the event that we have large differences (like 10%)
- See Joeys ppt on modeling intro for high visibility examples of the full drainage — Joey, can you send a link?
- note: this is not a knock on previous maps but they were zoomed in to show gage location and stream channel and didn’t tell us about other potential sources of error (like drainage errors)
- once we review all the full drainages, lets choose a threshold for automatic area-weighting (I like 10%) then reinvestigate those that are above our threshold
- keep in mind that there should be NO EXPECTATION that these watershed segments will coincide with gage locations. First off, USGS will do their own delineations that may or may not be shared with other agencies. Secondly, and perhaps most importantly, certain aspect of a river network may cause the modelers to make choices about segmentation that will appear odd — they usually have a good reason and if we find one of these we can speculate as to why and perhaps even ask the folks who created the model.
On Fri, Mar 16, 2018 at 12:40 PM Alyssa Ford notifications@github.com wrote:
There are differences in the drainage areas of USGS gages and the river segments we have in ArcGIS. I have not been able to compare these for every river segment yet, but so far the percent differences range from 0.26% to 46.29%. Rob and Joey have suggested area weighting to solve this problem.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/HARPgroup/cbp_wsm/issues/17, or mute the thread https://github.com/notifications/unsubscribe-auth/AEXAItDfZ5K-mbvdjc8fW0BHiTc9CgWHks5te-rqgaJpZM4SuGB2 .
--
Robert W. Burgholzer 'Making the simple complicated is commonplace; making the complicated simple, awesomely simple, that's creativity.' - Charles Mingus Athletics: http://athleticalgorithm.wordpress.com/ Science: http://robertwb.wordpress.com/ Wine: http://reesvineyard.wordpress.com/
alyssaf4
In the figure below, the Clinch River is flowing south. It leaves the river segment in the top left corner and enters the river segment in the bottom left corner. Between the time the river leaves the first river segment and the gage is located, a stream enters the river. This stream is not included in the drainage area of the river segment in the top left corner, but it is included in the drainage are of the gage, which is what causes the 11.97% error
There are differences in the drainage areas of USGS gages and the river segments we have in ArcGIS. I have not been able to compare these for every river segment yet, but so far the percent differences range from 0.26% to 46.29%. Rob and Joey have suggested area weighting to solve this problem.